implement table

pipestat/const.py

@@ -4,10 +4,12 @@
 REPORT_CMD = "report"
 INSPECT_CMD = "inspect"
 REMOVE_CMD = "remove"
+TABLE_CMD = "table"
 SUBPARSER_MSGS = {
     REPORT_CMD: "Report a result.",
     INSPECT_CMD: "Inspect a database.",
-    REMOVE_CMD: "Remove a result."
+    REMOVE_CMD: "Remove a result.",
+    TABLE_CMD: "Create a results table."
 }
 LIBS_BY_BACKEND = {"mongo": ["pymongo", "mongodict"]}
 CLASSES_BY_TYPE = {"integer": int, "float": float, "string": str,

pipestat/helpers.py

@@ -22,15 +22,27 @@ def build_argparser():
     # common arguments
     for cmd, desc in SUBPARSER_MSGS.items():
         sps[cmd] = subparsers.add_parser(cmd, description=desc, help=desc)
-        sps[cmd].add_argument("-d", "--database", required=True, type=str,
-            metavar="DB", help="database to store results in")
         sps[cmd].add_argument(
                 "-n", "--name", required=True, type=str, metavar="N",
                 help="name of the pipeline to report result for")
 
+    # remove, report and inspect
+    for cmd in [REMOVE_CMD, REPORT_CMD, INSPECT_CMD]:
+        sps[cmd].add_argument("-d", "--database", required=True, type=str,
+            metavar="DB", help="database to store results in")
+
+    # table
+    sps[TABLE_CMD].add_argument("-d", "--databases", required=True, nargs="+",
+                                metavar="DB", help="collection of file paths "
+                                                   "the results are stored in")
+
+    sps[TABLE_CMD].add_argument("-p", "--path", required=True, type=str,
+                                metavar="P",
+                                help="Path to the file where to store the result.")
+
     # remove and report
-    for subparser in [REMOVE_CMD, REPORT_CMD]:
-        sps[subparser].add_argument(
+    for cmd in [REMOVE_CMD, REPORT_CMD]:
+        sps[cmd].add_argument(
                 "-i", "--id", required=True, type=str, metavar="ID",
                 help="id of the result to report")
 

pipestat/pipestat.py

@@ -3,6 +3,7 @@
 import logging
 import logmuse
 import oyaml as yaml
+import pandas as pd
 
 from collections import Mapping
 
@@ -84,6 +85,15 @@ def cache(self):
         """
         return self._cache
 
+    @property
+    def table(self):
+        """
+        Tabular database
+
+        :return pandas.DataFrame: database in a table format
+        """
+        return pd.DataFrame.from_dict(self.database[self.name], orient="index")
+
     def report(self, id, type, value, cache=False, overwrite=False,
                strict_type=False):
         """
@@ -195,12 +205,16 @@ def main():
     global _LOGGER
     _LOGGER = logmuse.logger_via_cli(args, make_root=True)
     _LOGGER.debug("Args namespace:\n{}".format(args))
+    if args.command == TABLE_CMD:
+        pd.concat([PipeStatManager(d, args.name).table
+                   for d in args.databases]).to_csv(args.path)
+        sys.exit(0)
     psm = PipeStatManager(database=args.database, name=args.name)
     if args.command == REPORT_CMD:
         type = args.type
         value = args.value
         if args.type == "object" and os.path.exists(expandpath(value)):
-            # if reported type is object and value is and existing file path,
+            # if reported type is object and value is an existing file path,
             # try to load it as JSON. This way nested objects can be
             # reported using CLI
             from json import load

requirements/requirements-all.txt

@@ -1,3 +1,4 @@
 ubiquerg>=0.6.0-dev
 logmuse>=0.2.5
 oyaml
+pandas>=1.0.0